Another AI Rejection Post (5/30 Update)

An excerpt should still make sense within itself. These samples are hard to follow, just saying.
Don’t do that. That’s ridiculous to make that claim. They’re perfectly readable, in and of themselves. I just didn’t want to publish the whole thing.
 
You sound just a bit like an AI. You wouldn't mind me dunking you in the nearest lake to see if you short-circuit, would you?

:LOL:

I'm only joking.

I'm joking. But others, who say things like that, are not.

One of my students was accused of submitting AI-generated content in another class, and even threatened with disciplinary action for doing it. I found the accusation unlikely, and asked him to write a short essay for me while I watched. Had the essay checked, AFTER I PERSONALLY WATCHED HIM WRITE IT, and I was informed that an AI wrote it. So, that should prove that the test is flawed and can’t be used to accuse him, right?

The other teacher didn’t think so. She thought he deliberately wrote like a bot so that we won't notice it when he uses AI to write his regular school essays. She doesn't believe that any student can be as well-informed or erudite as he is, when fifteen seconds of speaking to him should confirm the fact.

So... I guess that's just where we're heading. If you're not a semi-literate ignoramus, you will automatically arouse the suspicion of those who are. If you can physically prove that you're not using an AI in your writing, they’ll suspect that you *are* the AI. What do you do then? Take the Voight-Kampff test?
Here is the main problem people don't get it. Students are not that good the way AI write. Nobody had any problems with grammar tools until chatgpt. When I was an computer science student, I got zero bonus mark for my good writing and zero penalty for bad writing. Nobody cares how you write your answer.

Teachers used to courage us to use Google and put more information in the assignments. Now due to chatgpt it's easy to find the information you need. So how can we stop students stop using AI? Make another AI who will detect is it human writing or not, bravo.

I know how this shit works and how it's made. The total system is broken. Yeap students are lazy but guess what so as the teachers. They are both using AI and complaining eachother.

Just change the evaluation system. Work hard and find a way to examine your students. How can you be so sure that AI writing detector giving you the right answer.

Just write a paragraph and put it in the best AI detector. Damn sure it'll come out as an AI writing more than 60% of times.
 
Ok. Now I'm really having trouble understanding how you mistyped "kids" as "kinder" in your story. It just seems a really weird mistake for a native English speaker to make.

I am trying to make it sound like a German native speaker speaking English.

That incudes using the German form of words in an English sentence.

It’s more accurate and more likely than saying “kids”.

It wasn’t a mistype, it was a deliberate editorial decision.

Where “kinder” was used, it was being said by a German native speaker.

For clarity:

Hans is Austrian. His first language is German. In speaking to Sam (British), he speaks English. His English is not perfect, so some German words filter in. Like kinder.

This is normal in the world.

That’s why I wrote it like that.

Deliberately.
 
The problem is they're using it the wrong way. AI writes like a human, not an alien. So, if you're good at writing, you'll be flagged most of the time. It's disgusting when they say my trashy writing is AI-written. You shouldn't use an AI detector here in the first place. To sort out bad stories, you need to read them yourself. If I use AI to write a story, then my stories will be so much better due to a good writing style. It still makes no sense who brought this idea. I mean, AI writes like us. We train them to write like this. They didn't innovate it, duh.

I completely agree, and have made this same retort on this forum many times. Every AI generator is trained with LLMs that are mined of documents on the internet. There are LLMs that are compiled of databases from this very site, which is the reason authors are being flagged.

I've been flagged multiple times, but more recently have been having more success getting stories through without problems.

My advice to help others is to keep a thesaurus handy and to do multiple checks and re-edits on stories, changing out adjectives. You want to try and avoid overusing specific words. It tends to pick up when the same words are used in descriptions for something. Another trick I found is to check your sentence structure, and try to avoid repeating the same sentences. I think the AI detection tools they are using are focused on picking up patterns in writing, which would explain why it's tripping on this.
 
I completely agree, and have made this same retort on this forum many times. Every AI generator is trained with LLMs that are mined of documents on the internet. There are LLMs that are compiled of databases from this very site, which is the reason authors are being flagged.

I've been flagged multiple times, but more recently have been having more success getting stories through without problems.

My advice to help others is to keep a thesaurus handy and to do multiple checks and re-edits on stories, changing out adjectives. You want to try and avoid overusing specific words. It tends to pick up when the same words are used in descriptions for something. Another trick I found is to check your sentence structure, and try to avoid repeating the same sentences. I think the AI detection tools they are using are focused on picking up patterns in writing, which would explain why it's tripping on this.
Had a bit of A look into the writer.com/ai-content-detector/ and what parts of my writing are setting off that detector.
For a specific section of 500 words it returned the following result after about 10 minutes of messing around trying to work out what triggered it I found one of the culprits to be the following.

"re-shifted" - with this word it returned a 98% human written.
If I change it to reshifted then the passage would return 99% human written.
but if I change the word to "shifted" then it would change my passage of text to 97% human written.

Hopefully someone finds this information useful.
 
Welp, 14 days after resubmitting, I am not surprised to have been rejected for AI again. Per their message, Literotica checked my story multiple times and it came back as the majority was written by AI. I honestly don't have the words to express the sheer absurdity of having a machine tell a human that their work was created by another machine - machines that have stolen our work in the first place. "We stole everything humanity has ever written and now we get to claim you're copying our homework."

So, again, I have basically no recourse. There is no way for me to convince them that my work is my own. Their suggestion is to work with an editor, but since the editors have no idea what's being flagged as AI either, how is that supposed to make a difference? Not to mention, this is not some paid gig. We all do this for free. I'll go through extra effort when submitting my non-erotic works to paying mags, but turning my fun side hobby into a massive chore just to try and circumvent an AI from falsely identifying my work as also being AI is not worth it. My work is at least as good as most other works here, and was good enough to get published the 1st time around. I'm not being rejected on quality or merit, but by the theft-robot.

I have yet to decide what I'll do next, but I am leaning towards just packing it up and saying fuck this place.

Also, if you're curious and want to judge for yourself, here's my 1st chapter, written in exactly the same style: https://literotica.com/s/immoral-counseling-ch-01
 
The point about 'AI' checkers is that they are black boxes. No one understands how they do whatever they do, so it is impossible to say 'change this' except in the most generic way. They are not AI either, but that's a different problem. I haven't submitted anything since before 'AI' became a thing, so never had it happen to me. I'm waiting to hear on the first couple of parts of a new series. It's never been near Grammarly or anything but a spell checker, but we'll see.
 
The point about 'AI' checkers is that they are black boxes. No one understands how they do whatever they do, so it is impossible to say 'change this' except in the most generic way. They are not AI either, but that's a different problem. I haven't submitted anything since before 'AI' became a thing, so never had it happen to me. I'm waiting to hear on the first couple of parts of a new series. It's never been near Grammarly or anything but a spell checker, but we'll see.
I agree. I hope it goes well for your current submissions and you don't face any issues. I've decided to give it one more try, almost as an experiment to figure out what's triggering the AI flag.
 
Oh my god, y'all, I might FINALLY be making progress. I don't want to jinx myself. My manuscript was rejected for a 3rd time today, but NOT for AI. My main female character is a graduating senior in high school and is 18 years old. I make this clear in chapter 1, but didn't think to do it in subsequent chapters. I added a note and resubmitted. I am crossing all my fingers and toes that it gets published successfully. If it does, I'll share what steps I took in hopes that it will help others who are having this AI issue. I know the vast majority of AH posters and regulars aren't encountering this, but for those who are, I hope it can help.
 
I updated this in the OP, too.

My previously rejected work is now pending publication for tomorrow. For over a month I've been dealing with this, so I am very happy to have finally succeeded in getting approved. I know that the AI rejection conversation is frustrating for a lot of you who aren't experiencing this. My hope is that by sharing what I did it gives insight into what's happening and what others can do if they get an AI rejection. By no means is this meant to be a comprehensive guide. This is just my experience, and others' may differ. I am also not putting this out there so that bad actors can circumvent detection. I just know from seeing other new writers that there are some of us who are trying everything to get approved and getting denied again and again (one person said he was denied 13! times).

So, that said, here's what I did.
  • My work is already 100% human written. If someone is using any kind of assist tool beyond a basic spell check, maybe stop doing that.
  • I ran my work through every single free online AI detector available to get an idea of what in my work might be tripping the sensor.
  • Changed basic, simple sentences to be more specific or colorful. Example: I changed "She didn't have time for [her friends]" to "She couldn't spare a second for them."
  • Used a lot more metaphorical language. Example: "He grunted and snorted like a pot-bellied pig." Or added vivid, descriptive words that aren't commonly used, such as "opalescent."
  • Rewrote dialogue to be more informal by removing unnecessary nouns and verbs. Added more unusual contractions. Example: "What're" instead of "What are." Also made dialogue more colorful where possible.
  • Paid closer attention to varying sentence length and structure. I have a problem with always wanting to write long, multi-clause sentences.
  • Broke up or rewrote simple sentences joined by ", and" because that seemed to get flagged by the AI detectors a lot.
  • Disrespected the rules of grammar. Inserted tons of incomplete clauses and sentences where it made sense. Turned full sentences into fragments to convey the same information. Improperly using the -ing form of a verb. Example: "Clamping his lips over her nipple, sucking it hard before gently raking his teeth across the stiff bud."
The big things seemed to be, write colorfully, break grammar rules often, and don't overuse common or cliche phrases.

I genuinely hope this helps, and I hope this is the last I'll speak on this subject. I've got a lot more stories I want to put out, so I'm hoping to not have to go through all this for every single one.
 
Congratulations! It must be a huge relief. I can't imagine the frustration you must have felt, or how much energy it must have cost to keep going at it again and again.

I hope the readers are appreciative!
Thank you! I've submitted non-erotic works to major publications, so I'm used to getting rejected A LOT. So, you know, that helped. XDDD

But yes, HUGE relief. Now I can focus all that energy on new chapters instead of rewriting one chapter over and over.
 
I updated this in the OP, too.

My previously rejected work is now pending publication for tomorrow. For over a month I've been dealing with this, so I am very happy to have finally succeeded in getting approved. I know that the AI rejection conversation is frustrating for a lot of you who aren't experiencing this. My hope is that by sharing what I did it gives insight into what's happening and what others can do if they get an AI rejection. By no means is this meant to be a comprehensive guide. This is just my experience, and others' may differ. I am also not putting this out there so that bad actors can circumvent detection. I just know from seeing other new writers that there are some of us who are trying everything to get approved and getting denied again and again (one person said he was denied 13! times).

So, that said, here's what I did.
  • My work is already 100% human written. If someone is using any kind of assist tool beyond a basic spell check, maybe stop doing that.
  • I ran my work through every single free online AI detector available to get an idea of what in my work might be tripping the sensor.
  • Changed basic, simple sentences to be more specific or colorful. Example: I changed "She didn't have time for [her friends]" to "She couldn't spare a second for them."
  • Used a lot more metaphorical language. Example: "He grunted and snorted like a pot-bellied pig." Or added vivid, descriptive words that aren't commonly used, such as "opalescent."
  • Rewrote dialogue to be more informal by removing unnecessary nouns and verbs. Added more unusual contractions. Example: "What're" instead of "What are." Also made dialogue more colorful where possible.
  • Paid closer attention to varying sentence length and structure. I have a problem with always wanting to write long, multi-clause sentences.
  • Broke up or rewrote simple sentences joined by ", and" because that seemed to get flagged by the AI detectors a lot.
  • Disrespected the rules of grammar. Inserted tons of incomplete clauses and sentences where it made sense. Turned full sentences into fragments to convey the same information. Improperly using the -ing form of a verb. Example: "Clamping his lips over her nipple, sucking it hard before gently raking his teeth across the stiff bud."
The big things seemed to be, write colorfully, break grammar rules often, and don't overuse common or cliche phrases.

I genuinely hope this helps, and I hope this is the last I'll speak on this subject. I've got a lot more stories I want to put out, so I'm hoping to not have to go through all this for every single one.
You essentially rewrote your piece to satisfy AI, allowing it to interfere with your work. Personally, I'd rather face rejection than change a single syllable.
 
The big things seemed to be, write colorfully, break grammar rules often, and don't overuse common or cliche phrases.
Other than maybe the grammar thing, this sounds like sensible writing advice in general.

Do you feel like your story is/reads better now, after making those changes?
Or was it more along the lines of what @Plathfan is alluding to, i.e., needless interference that only served to satisfy arbitrary requirements that don't necessarily agree with your artistic vision?
 
(one person said he was denied 13! times).
Ayy, that was me 😂 I did manage to get something through back in March (which is, ironically, my most favourited/read story sitting at a 4.7 rating) but it was a world away from my usual style. I did the same thing as you and changed the way I wrote just to see what would happen.

Since then, I haven't even tried to submit anything. I've had ideas, but I don't want to get stuck in the cycle of rejection all over again. It's not worth the hassle.
 
You essentially rewrote your piece to satisfy AI, allowing it to interfere with your work. Personally, I'd rather face rejection than change a single syllable.
On the other hand, it sounds like he followed what would, two or three years ago, have been typical writing advice - writing something that's lively and interesting, not bland, cliched and monotonous.
 
To address the questions/comments from @Plathfan @TheLobster & @ElectricBlue -

I weighed my options and made a decision to revisit my story, but I didn't completely change who I am, or change my voice. Within a certain margin, I didn't mind making changes to what was there. In the end, I was still actively deciding what and how to rewrite and what to leave the same. The way I rewrote those passages is still in line with my writing style, just /more/ if that makes sense. And some of what I changed I do like much better, so there's that. While it was a compromise in a sense, it's still my work and my voice, so I'm proud of it. If I couldn't have kept it like that, I would've given up. Thankfully, it didn't come to that.

I also treated it as an experiment. Could I overcome this issue and how? I wanted to know. It was worth it for me to find that out, and especially to know if what I had to was within reasonable bounds.

I will say, I think what I did does go beyond just general good writing advice, just because I think the way these detectors flag straightforward prose encourages overwriting. Like, not every sentence has to be doing the most. A flat, simple sentence can get the job done, and creates a nice balance and contrast with more flowery, dramatic prose. I kept it within a margin I am comfortable with, but absent this exercise, I would've been more conservative with my writing overall. So I guess you could say it interfered with my artistic vision, but overall, pretty minor. There's a few lines of prose, < 1%, that I probably wouldn't have written otherwise, but that's it.
 
To address the questions/comments from @Plathfan @TheLobster & @ElectricBlue -

I weighed my options and made a decision to revisit my story, but I didn't completely change who I am, or change my voice. Within a certain margin, I didn't mind making changes to what was there. In the end, I was still actively deciding what and how to rewrite and what to leave the same. The way I rewrote those passages is still in line with my writing style, just /more/ if that makes sense. And some of what I changed I do like much better, so there's that. While it was a compromise in a sense, it's still my work and my voice, so I'm proud of it. If I couldn't have kept it like that, I would've given up. Thankfully, it didn't come to that.

I also treated it as an experiment. Could I overcome this issue and how? I wanted to know. It was worth it for me to find that out, and especially to know if what I had to was within reasonable bounds.

I will say, I think what I did does go beyond just general good writing advice, just because I think the way these detectors flag straightforward prose encourages overwriting. Like, not every sentence has to be doing the most. A flat, simple sentence can get the job done, and creates a nice balance and contrast with more flowery, dramatic prose. I kept it within a margin I am comfortable with, but absent this exercise, I would've been more conservative with my writing overall. So I guess you could say it interfered with my artistic vision, but overall, pretty minor. There's a few lines of prose, < 1%, that I probably wouldn't have written otherwise, but that's it.
I ended up just resubmitting, this time it was my document instead of copying the text. Tracked changes were on and in the notes section I put a declaration the work was my own. That got approved.
 
Back
Top